Signed-off-by: Jessica Yuen <im.jessicayuen@gmail.com>

Signed-off-by: Jessica Yuen <im.jessicayuen@gmail.com>

Signed-off-by: kyamazawa <yamazawa@supersoftware.co.jp>

Currently, if we run `param set guestbook-ui foo-bar x`,

The jsonnet snippet will write: `foo-bar: "x"`.

However, this is incorrect jsonnet syntax. This commit will fix this
issue, to instead write `"foo-bar": "x"`.

Also updates the `Get...` Param functions to return non-quoted
identifiers for prettier display.

Signed-off-by: Jessica Yuen <im.jessicayuen@gmail.com>

As part of the environments feature, we want to be able to deploy to a
specific cluster given the environment's URI. We cross-check against
kubecfg for this URI's location. For fail-safe comparison, we probably
want to normalize these URIs.

A bug existed where if `param set` was called for the same param twice,
a duplicate entry was created for the component. This only occured where
the component name contained a special character.

Historically, every time we've needed to write out something tabular,
we've hand-rolled an ad hoc padded row writer. Yes, in each case. Such
is #startuplife.

The third time we did this we wrote this with bugs, and so now we are
rewriting it in a general, tested function, and transitioning all places
where we call this to use this function.

Currently, if a component name contains a special character, ex:
foo-bar, this translates to the jsonnet identifier: foo-bar, which is
invalid syntax.

This change will quote component names where there are special
characters.

This change enables the user to diff between two environments that are
either local or remote.

i.e.,

`kubecfg diff local:dev local:prod` will diff between the expanded
templates for each environment on disk.

`kubecfg diff remote:dev remote:prod` will diff between two remote
environment clusters. It does this by first expanding the component
templates of each environment. Then, the live objects are fetched from
each of the clusters and the diff is performed against the live objects.

`kubecfg diff local:dev remote:prod` is also an option. This will diff
between the expanded templates for 'dev' on disk and the live objects
on 'prod's server.

Fetches schema from server without caching to disk.  Caching / offline
operation will be added in a followup PR.

- Move fqname to utils
- Add ResourceNameFor and use it to get a lowercased/pluralised
  version of the resource name

Fixes #15

jsonnet's `std.toString()` returns "compact" JSON.  Sometimes this is
not appropriate, particularly when generating large output that might
need to be viewed/debugged by a human.

This change implements manifestYaml and manifestJson functions, which
generate YAML and "pretty" JSON respectively.

A slight implementation wrinkle is that libjsonnet currently only
supports passing primitive (scalar) types to native functions, so
these arguments need to be JSON-serialised/unserialised across the
native function call boundary.

Annoyingly, delete options `OrphanDependents` and `PropagationPolicy`
are mutually exclusive.  This change compares the k8s server version
to see if the server is going to look for the newer PropagationPolicy.

Fixes a78be141 (aka PR #59)

Improve interactive output experience by switching to a logging
library (logrus) that allows customising the output format.

This also removes the glog command line flags.  Replaced with a
simpler `--verbose`/`-v` option:
- quiet by default (warnings and errors only)
- `-v` for progress info
- `-vv` for debug

Fixes #34

Fixes #35

- `gofmt -s utils/registry_test.go`
- Fix missing `if err != nil` check in `cmd/delete.go` (a bug, but not critical)
- Add goreportcard button to README

- `escapeStringRegex`
- `regexMatch`
- `regexSubst`

They do what you expect, as implemented by the golang `regexp` package.

Add a new native function that resolves docker image names into more
specific forms.  In particular, it can look up a docker registry and
convert image:tag to image@digest at jsonnet-eval time.

Limitations: Does not currently support private docker
registries (that require authentication).

Controlled via two new command line flags:
- `--resolve-images` Change implementation of resolveImage native
  function. One of: noop, registry (default "noop")
- `--resolve-images-error` Action when resolveImage fails. One of
  ignore,warn,error (default "warn")

Note in particular that the defaults will *not* do remote registry
lookups, and will only add an explicit ":latest" tag where no tag was
given.

Fixes #13

Fixes #12

Uses github.com/yudai/gojsondiff for the heavy lifting.

This removes all objects listed in config files (and dependents) -
effectively reversing an `update --create`.

Intended for short-lived "try this out; now clean up" workflows.

Fixes #14

There is some variation in jsonnet "result" conventions.  This change
assumes the top level is a JSON object, and then walks down looking
for objects with `apiVersion` & `kind` fields.

While walking down, the current code will expand JSON arrays, and
object values.

In particular, this should correctly accept:
- single-value JSON files
- arrays (jsonnet --yaml-stream)
- "filename->object" objects (jsonnet --multi)

(Note that any v1.Lists encountered are still expanded in a
second-pass, as before)

The goal is to make a best-effort attempt at reducing the number of
"crash-restart" loops required to bring up a group of interdependent
resources.

The current implementation is very simple and just sorts into 3 tiers:
- Namespace, ThirdPartyResource, StorageClass
- everything else
- Pods or similar (Pod/Job/Deployment/DaemonSet/StatefulSet)

- Support the standard client-go/kubectl flags
- Switch object types to use client-go rather than apimachinery
- Implement update subcommand

File format is recognised via simple file extension check.

Any "List" objects found will be expanded into their constituent
items, so single-element formats like json and jsonnet can use a
v1.List to capture multiple resource objects.